How AI Contributes to the Audiobook Industry Boom
Audiobook sales have been skyrocketing in recent years with Audio Publisher Association (APA) reporting that in 2018, US publishers sold around $940 million audiobooks, a 24.5% increase over 2017. APA also notes that 91.4% of 2018’s revenue came from digital format sales. This shows how enormous digital media’s impact is on the audiobook industry growth.
Thanks to platforms like Audible, audiobooks are now more accessible. Users can listen to an audiobook on their phone, connect it to their car, listen to it on their tablet, laptop, and smart speakers. It will even automatically remember where the user is in the audiobook and sync through all their devices.
The Use of AI in the Audiobook Industry Today
Recently Audible got in trouble for its AI-powered feature. Five major publishers of the US are suing Audible for its Captions feature. The feature that the company announced this summer uses machine learning to transcribe audiobooks into text. Because audiobook licenses are separate from ebook licenses, publishers are claiming this is a copyright infringement. This use case of AI in audiobooks seems controversial considering copyright limitations, but it might become a legal part of audiobook platforms in the near future if two parties are able to strike a copyright deal.
The subtitle feature might be especially useful for users who are using audiobooks to learn a language. The language learning sector has humongous revenues itself, with English learning global market size expected to reach $22000 million by 2024, according to Reuters.
Another way Audible utilizes AI is by its personalization of audiobook recommendations. Just like every major streaming platform out there, the platform uses user’s audiobook listening history to offer better suggestions, so that they always have something to listen to and keep their subscription for as long as possible.
While Netflix goes as far as using auto-generated and personalized thumbnails for its movies and shows, we don’t see this use case in audiobook platforms just yet. But who knows, maybe that’s in the works too for people who judge a book by its cover.
New Opportunities for Audiobooks, Brought to You by AI
· Automating Audiobook Narration
While AI is far from matching the quality of human performance just yet, it already is able to imitate human speech. AI helps advance text-to-speech technologies sound like natural voice recordings. NaturalReader, a free online text-to-speech converter, is one of those tools that does a pretty good job at making natural-sounding voice recordings. Animaker even lets users customize the AI-generated speech recording by inserting brief pauses and breathes to make it sound like an actual person speaking.
Making AI text-to-speech sound like a human has been tackled by tech giants for a while now. Apple, Google, Microsoft, and Amazon have all been working to make their AI personal assistants’ voices sound more human. As stated in Scientific American’s article, these systems work by connecting words and phrases from prerecorded files. But that’s not enough to make them sound natural. Without the nuances of human speech, such as pauses and breathes that we do while we’re speaking, these systems sound robotic. This is where deep learning enters the scene. By analyzing different human speech recordings, AI can learn to mimic nuances that make speech sound more natural.
· Creating Short Summaries of Audiobooks
With apps like Blinkist, a so-called condensed reading platform, getting more and more popular, a new opportunity opens for AI in the audiobook market. With AI, platforms like Amazon and Audible can use AI to create short summaries of books with essential key takeaways for people who don’t want to invest hours into a book. The existence of platforms like Blinkist, which recently raised $18.8 million in funding, shows that there is a demand for short versions of books, and it can be an excellent opportunity for audiobooks.
Not surprisingly, there are already tools that attempt to do this using artificial intelligence. AISummarizer is a free online tool that you can paste your text in and get a summarized version. It lets the user choose how much the text should be reduced based on percentage.
· Narrator Accent & Voice Adjustment
TechCrunch reports that recently, Amazon applied for a patent for an audio system that can detect the accent of a speaker and adjust it to the accent of the listener. The technology has a potential for the audiobook industry as well. Imagine listening to an audiobook and having a feature where you can select the accent of the narrator. As far fetched as it sounds, this feature can be handy for those cases when you found an interesting audiobook, but the narrator has an accent that makes it harder for you to listen to it.
Or what if you want a familiar voice reading the book to you? There’s an AI solution to that as well. Bloomberg recently interviewed the team of Lyrebird, a startup that uses AI to clone human voices. To achieve that, they are using the above-mentioned deep learning technique to analyze a one-minute voice recording and create an AI version of it that can be manipulated to speak any kind of text you feed to it. It even preserves accents. In the wrong hands, a tool like this can have destructive consequences, while on the other hand, it can have sentimental uses too, like getting nostalgic and listening to your mother read you a fairytale.
Audiobooks are on the rise right now, and soon competitors will want to stand out with unique features. Whether it’s presenting new ways to enjoy audiobooks or just enhancing the production process, AI has a lot to offer to this industry. For now, we just have to wait and see what’s next for audiobooks.
By Max Kalmykov
VP, Media & Entertainment at DataArt
